Doctorante, Polytech Annecy-Chambéry, LISTIC.
Contact
Courriel : Sabrine.bouaziz@univ-smb.fr
Adresse : LISTIC, 5 chemin de bellevue, CS80439, 74 944 Annecy Cedex
Bureau : A224
Thèse
Theme : AFuTé
Sujet : XAI (eXplainable AI) for rockfalls prevision.
Résumé :
This thesis aims to enhance the management of the risk associated with rockfall events, which are rare but devastating occurrences leading to severe consequences such as loss of human lives and damage to infrastructure. Current challenges include the difficulty in predicting the location and volume of future rockfalls to design appropriate protective measures, while considering the financial constraints of municipalities and private operators. Furthermore, risk management requires a dynamic estimation of the probability of rockfall occurrence based on weather conditions, involving high socio-economic costs.
The thesis proposes overcoming these challenges by leveraging Artificial Intelligence (AI) methods to better understand the behavior of cliffs under the influence of weather conditions and develop effective predictive models. The objective is to replace expertise-based approaches with more reliable quantitative methods. To ensure the relevance of the results in the context of public decision-making, the study emphasizes the need to provide understandable explanations for decisions generated by AI models.
Mots-clefs : Artificial Intelligence, Explainability, Deep learning, Data Mining, Rockfall, Risk
Directeur de thèse : David Amitrano (IsTerre)
Co-directeur : Christophe Lin-Kwong-Chon (LISTIC)
Début de la thèse : 08/04/2024
Ecole doctorale :
Sciences, Engineering, Environment (SIE)